Imagine a world where you not only know an individual’s most preferred movie choice, but also how likely they are to see it? Movio’s second insights-driven module, Audience Insights, allows any cinema marketer to do just that.

With the introduction of Movio’s proprietary Propensity Algorithm™, Audience Insights is set to revolutionize the way cinema marketers engage with moviegoers. In this blog post, we explain how propensity has been introduced to our Insights modules, and what it can do for movie marketing.

The power of propensity

The ability to know what the audience distribution looks like for any given week is the holy grail for movie marketers and moviegoers alike.

Exhibitors can:

• Visualize the distribution of members for each movie screening to focus campaign tactics and balance marketing efforts for the coming week based on the likely audience size for movies ‘Now Showing’ and ‘Coming Soon’.
• Identify the best content to feature in emails and newsletters by easily analyzing the demand for each movie showing in circuit.
• Modify incentives by calibrating offers to each individual's likelihood of seeing a movie.
• Optimize scheduling by supplementing the experience of film buyers by illustrating the likely share of admissions for the coming week.
• Improve targeting by using Movio Views to create a lifestyle or behavioral segment and find the ideal movies for those members.

Moviegoers will:

• Receive the most relevant content in the communications sent to them, personalizing each customer’s experience.
• Be matched with their most likely movie choice for what is currently showing in cinema.
• Benefit from potential incentives that can make a difference to them seeing the movie being promoted to them.

Advanced targeting with Audience Insights

Once we proved that propensity can successfully determine the likelihood of someone watching a movie based on their historic moviegoing behavior, it was simply a matter of defining how we could put it to use in Movio Cinema.

Our first implementation of the Propensity Algorithm is in our latest module, Audience Insights. It provides Movio Cinema users with the ability to easily identify the most likely movie choice for any loyalty member from the movies currently showing in their cinemas. Being able to automatically and instantaneously determine this will revolutionize the way marketers promote the most relevant content to their loyalty members.

Powering up Movie Insights

For movies that are yet to be released, but have presale tickets available for purchase, users can create ‘lookalike’ audiences by modeling comparable titles (comp titles). These models can then be used to find the best audience for upcoming movies until enough transactional data is available for a particular title. Each model has a target movie, and one or more comp titles recommended by Movio’s Similarity Algorithm™, originally introduced with the launch of the Movie Insights module in 2017.

As Matthew Liebmann stated in our recent white paper, The New Wave of Movie Marketing, “Movio’s Similarity Algorithm has been embraced by exhibitors, studios and other cinema industry stakeholders with demonstrable success. However, from our perspective, it is still a binary instrument: either you’re in the group or you’re not.” We soon realized that by introducing propensity into Movio Cinema, we could also take the Movie Insights module to a new level. Rather than simply questioning whether or not a member should see a particular movie, users can now determine how likely they are to see it.

Movie Insights splits the audience into three main segments based on their likelihood to see a movie:

• Most likely - 5 times more likely to see the promoted movies than the average frequency of all of all active members in the exhibitor’s database.
• Likely - 3-5 times more likely to visit than the average frequency of all active members in the exhibitor’s database.
• Less likely - 1-3 times more likely to visit than the average frequency of all active members in the exhibitor’s database.

These insights allow users to create more tailored messaging, and the ability to match the value of incentives to each moviegoer’s individual propensity.

Testing Movio’s Propensity Algorithm

In order to illustrate that the algorithm produces highly targeted audiences and relevant recommendations, Movio’s Data Science team ran a series of experiments to test which movies the Audience Insights module would recommend to members during the the first week of Black Panther. The team looked at 10 movies which were in cinemas at the start of that week.

Using multiple years of moviegoer data prior to the start of that target week, the Propensity Algorithm™ ranked these 10 movies for each member, from most likely to least likely.  

Chart A shows the predicted audience distribution for all members. It predicts what an imaginary distribution will look like if all members were to watch their top one movie based on Audience Insights’ recommendations.

It's important to note that creating a good comp model of a new release is crucial for getting a reasonable audience distribution prediction. Diversity is as important as accuracy at this stage as we often want to boost the visitation rates of ‘smaller’ movies. A more diverse distribution will give the ‘smaller’ movies more campaign recipients. You can see this chart as a recommendation of optimal campaign resource and effort assignment.

Chart B shows the actual audience distribution of visiting members.

For the opening week of Black Panther (Feb 16, 2018), when investigating the moviegoing behavior of visiting members, we found that:

• 55% of visiting members saw their top 1 recommendation.
• 68% of visiting members saw a movie in their top 2 recommendations.
• 78% of visiting members saw a movie in their top 3 recommendations.

It’s worth noting that it is virtually impossible to make movie recommendations with 100% accuracy due to a range of environmental issues that influence the movie ultimately selected by a member, including the type of cinemagoing occasion, who they intend to go with and the the availability of convenient showtimes. And of course, in some instances, the member doesn’t know they will like our recommended movie more than the one they ultimately watched!

The team continued to run similar experiments for multiple weeks. The results that were collected from each week are consistent and show that, at a macro level, the Audience Insights module provides valuable marketing insights by precisely predicting the audience propensity distribution of movies selected each week. At a micro level, the personalized recommendations are consistently accurate and relevant.

The machine learning algorithm explained

‘Machine learning’ may just sound like the latest buzzword, but it has been used for a long time. It has evolved immensely over time and has proven to be more than capable of automatically performing complex computations in order to find patterns in big data, in a fast and repetitive fashion. With technological advancements, this automated way to search and process predictive insights from huge volumes of data became more accurate, faster, extremely powerful and cheaper to implement.

Movio’s Propensity Algorithm uses machine learning based propensity models to determine the likelihood of someone watching a movie using individual and collective behavioral data. Here is a simple way to explain this:

• Sam watched Blade Runner 2049, Thor: Ragnarok, Black Panther and Annihilation.
• These four movies were also watched by members who also purchased tickets to Ready Player One.
• Therefore, Sam is highly likely to watch Ready Player One.

Audience Insight’s machine learning algorithm uses Sam’s and other member’s moviegoing histories to determine the likelihood of Sam to watch Ready Player One.

By using machine learning techniques, Movio aims to help marketers focus on the important jobs in their day-to-day tasks, rather than having to do these computations manually.

The bottom line

Audience Insights provides our users with the most powerful tool in the industry for advanced segmentation and targeting, allowing them to fully focus their expertise on creative compelling content. It enables movie marketers to automatically identify the most relevant movie for each moviegoer and tailor their marketing efforts and incentives accordingly: an ultimate transformation of the way movies are marketed.